Year: 2023
Type: dissertation
Abstract: The rapid development of the Internet of Things (IoT) has created a huge and complex network of interlinked equipment generating massive amounts of data. To efficiently analyze this data for a variety of applications, we propose an intelligent approach that utilizes multi-source and heterogeneous data from IoT devices. Our approach incorporates centralized and distributed learning techniques and is implemented through a set of secure, flexible, and scalable microservices. We divide the data analytics functionality into containerized microservices and train analytics models using AI methods before deploying them on edge computing nodes. We also utilize transfer learning and ensemble learning techniques to enhance model generalization and prediction accuracy. The adoption of a microservices-based architecture offers several advantages, including scalability, flexibility, reliability, modularity, and integration. This simplifies the development, implementation, and management of intricate machine learning and deep learning systems compared to conventional monolithic architectures. Our approach has significant implications for the IoT industry, providing a powerful tool for unlocking valuable insights from IoT data. We validate our approach through a set of IoT case studies and comparative studies, demonstrating its effectiveness and practicality in real-world scenarios. Overall, our suggested approach offers a robust and effective solution for analyzing IoT data and deriving important insights that can be used in a wide range of use cases.
Source: HAL
Author: Ben Atitallah, Safa
Institution: Manouba University
Language: English
Topic: IoT and Edge/Fog Computing
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4392224917
